Information Technology Reference
In-Depth Information
Table 1.
General-purpose secure multicast schemes comparison
Storage overhead
Communications overhead
Key Server Member
Join
Leave
Multicast
Unicast
Multicast
GKMP [5], 1997
2
b
2
b
2
b
2
b
(
n −
1)
b
Clusters [21], 1999
c
d
d
−
1
+
c
log
d
(
c
)+2
c −
1+
dlog
d
(
c
log
d
(
c
)+2
c −
1+
dlog
d
(
c
)
)
LKH [6][7], 1999
(2
n −
1)
b
(
h
+1)
b
(2
h −
1)
b
(
h
+1)
b
2
hb
ELK [15], 2001
(2
n −
1)
b
(
h
+1)
b
0
(
h
+1)
b
h
(
b
1
+
b
2
)
LKH++ [11], 2002
(2
n −
1)
b
(
h
+1)
b
b
+
log
2
n
(
h
+1)
b
log
2
n
+(
h −
1)
b
OFT [13], 2003
(2
n −
1)
b
(
h
+1)
b
(
h
+1)
b
(
h
+1)
b
(
h
+1)
b
(
dn
−
1)
b
d
−
1
SKD [12], 2009
hb
h
hb
(
d −
1)
hb
EGK [20], 2010
log
2
n
log
2
n
b
2
b
log
2
n
stateless but its computation complexity becomes excessive when increasing the
number of members. To alleviate that problem, Scheikl et al. combine it with
the LKH approach, thus obtaining a stateful protocol [24].
Most of the protocols reviewed above can handle large, dynamic audiences
in many multicast services that demand privacy. Regarding security, they are
mainly collusion free except for OFT and FT. Finally, and regarding reliability,
we must remark that LKH was introduced in the late 90's, a time when the main
part of communications were held on reliable links. That is the main reason why
these schemes normally do not address reliability issues and therefore they are
statefull and not self-healing. Probably ELK is the most reliable protocol in this
family. Table 1 shows a comparison in terms of storage and communication costs
of the most important schemes along with recent proposals (SKD and EGK).
Data are expressed in bits. There are small but subtle variations in the results
shown. It can be seen that older schemes focus on reducing communications in
re-key operations, while acquiescing in linear storage needs. More recent pro-
posals focus on the latter, given that acceptable bandwidth usage results were
already obtained. Other reasons for storage reduction are the popularization of
smart devices (with low storage capabilities) and the ever increasing audiences as
multimedia multicast services become more and more popular. As a last remark
we note that statistical improvements can be applied to member arrangement
in tree-based schemes in order to gain eciency and scalability [25].
3 Multi-group Schemes
There exist scenarios in which several, different information channels are en-
crypted separately and reach different, not disjoint groups of members. Typical
examples are multimedia platforms with several pay-per-view channels and com-
munications in hierarchically managed networks. Schemes shown next can be
seen as an extension of the tree approach: multiple trees are built from a single,
global set of leaves, thus obtaining several roots. Figure 1(b) shows an arrange-
ment example. Since a single member may now belong to more than one group,
her key path includes all keys from her leaf to the different roots she is connected
to, therefore re-key operations will normally affect more than one tree (however,
note that not all users are always connected to all roots). The pioneer proposal,
Search WWH ::
Custom Search